# Phase 8 Summary: Section 9.2 Hypothesis Tests

*Three mechanisms for why Z₁'s CA coefficient drops 9:1 after capital account opening in transition economies.*

## Test 1: Disruption of Structural Confounders

*Pre-opening, Z₁ proxies for Soviet-era structural characteristics. Opening breaks this correlation.*

- **1a (Partial R²)**: Pre R² = 0.6496, Post R² = 0.0616 (ratio = 10.55×)
- **1b (Mediation)**: Pre attenuation = 87.8%, Post attenuation = -9.6%
- **1c (Oster δ)**: δ_pre = 0.721, δ_post = -0.249

**Verdict: SUPPORTED** — Majority of tests consistent with confounders hypothesis.

## Test 2: Countervailing Capital Flows

*Opening enables both inflows and outflows that offset the lifecycle pattern in net terms.*

- **2a (Gross flows)**: 1/3 asset components significant, 1/3 liability components significant post-opening
  - No clear offsetting pattern in gross flows
- **2b (CA components)**: Savings and investment channels tested
- **2c (Capital flight)**: Δ FX reserves proxy tested

**Verdict: NOT SUPPORTED** — No clear offsetting pattern in gross flows.

## Test 3: Composition Effects Across Cohorts

*Early openers (Baltics) vs late openers (Central Asia) have different trajectories; pooling obscures patterns.*

- **Late openers**: Z₁ pre = 199.403, post = -1.382, ratio = -0.007
- **Continuous fade**: Z₁×event_time = 0.383**
- **Discrete break**: Z₁×post = -0.304

**Verdict: INCONCLUSIVE** — Insufficient data for both cohorts.

---

## Overall Assessment

| Mechanism | Key Evidence | Verdict |
|:---|:---|:---|
| 1. Structural confounders | Partial R², mediation, Oster bounds | See above |
| 2. Countervailing flows | Gross flow decomposition | See above |
| 3. Composition effects | Cohort-specific attenuation, trajectory divergence | See above |

*All tests: PanelGLS with entity and time effects. Transition economy openers only.*